J11-1005 our word segmentation model the word-based approach . In fact , word-based segmentors
C96-1037 sentences for training . By using a word-based approach , less frequent words or words
J14-4003 therefore refers to O&R 's word-based approach . Table 5 shows that on our data
E95-1014 evidence suggest a limitation of our word-based approach . Multi-word phrases , such as
D14-1003 net - works . The first one is a word-based approach using word alignments . Second
D14-1125 effect ) and Null ) . A purely word-based approach is blind to these cases . 3 Related
J11-1005 and Clark ( 2007 ) we proposed a word-based approach to segmentation , which provides
E09-1074 2007 ) . For the syntax-based and word-based approaches , we only took into account features
E12-1005 not outperform various context word-based approaches in two phrase similarity tasks
C04-1067 show the trade-off between the word-based approach and the character-based approach
D13-1119 character-based approach . The word-based approach searches for all possible segmentations
J90-2002 regularities that our current word-based approach must overlook . Finally , we
H91-1026 correspondences . The advantage of the word-based approach becomes important for complicated
D15-1295 window . In contrast to previous word-based approaches , our model induces vector representations
D14-1003 wordbased and phrase-based . The word-based approach assumes one-to-one aligned source
J11-1005 show that the flexibility of the word-based approach , enabled by our general framework
J11-1005 represent its segmentation . Our word-based approach does not map the segmentation
J11-1005 the framework can accommodate a word-based approach , rather than the standard and
D13-1012 mitigating the problem relative to word-based approaches . Using the TIRE model , we can
D15-1080 p-value > 0.05 ) , the simple word-based approach does have a slight edge over
hide detail